[RFC-0011] PoS inflation adjustment

Let’s ignore the scenario that I described where you try to take advantage of the single-block calculation time, as i dont think that is what you are concerned about. Instead, let’s imagine a scenario where 33% of the network is minting on a regular basis without you, making nInflationAdjustment=3, but with some random deviation. You have say 1% of the supply, and you are waiting to mint once or twice a year, and you try to time it for maximum profit. The effect your 1% will have on the nInflationAdjustment is somewhat negligible, as it is only 1/33 of the current participation, so you aren’t really concerned with your own participation affecting the system. Rather, you are aiming to speculate for moments when the network is at its lowest participation such that you get the highest reward. You analyze the yearly oscillations and find that, say, May and November are the times when you could make the most. Normally, you would mint in February and October, but you hold off till the mint difficulty is lower to participate.

My question is, what is wrong with this? You are now seeking out moments when the PoS difficulty is lowest in order to participate as a good actor. If anything, this makes the network over all more secure. You are not minting fewer times than you normally would, as you are absolutely convicted to only mint twice a year anyway.

In addition to all this, the compounding interest is increased by a significant amount as the % goes up from 1 to 5, which would discourage these tactics and cause people to mint more often.

1 Like

Yes, those seasonal oscillations are exactly what I am worried about.

This seeking out of moments when the PoS difficulty is lowest is itself the problem, because the coins are not minting during this time. This reduces the security of the network and leads to a decline of the PoS difficulty (self-fulfilling prophecy). And keep in mind that the standard assumption is that the vast majority of actors are selfish.

Yes, the compound interest will maybe safe the dynamic system that you want to introduce.
The problem with dynamic systems is that they can often be gamed. There could also be other attack vectors, which we didn’t think of yet.

I remember that in the past, a constant stake reward was discussed to encourage minting participation, but the idea was dropped. I think it was because selfish actors would split their stakes into a huge number of tiny outputs, so that after each successful stake, only a small amount of coindays would be lost. Is that correct?

I am asking because I am thinking about a somewhat similar idea. It would increase the reward% like in Nagalim’s proposal, but as a constant, and with a maximum reward cap, so that the annual PoS inflation does not exceed 1%.
The difficulty lies in finding the right reward cap. A conservative way would be to calculate the reward cap as:

Number of coins in circulation x 0.01 / 52560*

*number of annual PoS blocks

This would result in a maximum reward of about 4.79 PPC per block, which would increase by a tiny bit with every block. Annual PoS inflation would reach 1% only if every PoS block would hit the maximum reward. In reality, many stakes would mint before hitting the reward cap, so it could be set to a higher number, something between 5 and 10 PPC. But there is probably no way to determine the “right” cap with a fixed formula.

In the described scenario, this minter was already a seasonal minter. This scenario is one where the seasonal minter is damping seasonal oscillations that already exist, not amplifying then.

Again, the assumption in this scenario is that this actor will only mint twice a year, so they have not reduced the frequency of minting at all. Rather they are simply adjusting the timing of their minting to coincide with when the network has low difficulty. You seem to want to hold constant the frequency of minting, but then state that this decreases the frequency of minting, which is a contradiction.

This is mentioned in the proposal under the ‘alternatives’ section.

This is named the ‘stake grind’ in the ‘conventions’ section. It has more general implications for any proposal involving the reward.

This will result in a stake grind. The question will be what the optimal stake size is to get close to but not exceed the maximum reward cap, and everyone will break their stake into those chunk sizes. This will bloat the UTXO table and make staking less weildly.

Nagalim,

Lets halt the discussion about RFC11 here. Ultimately, it’s about human behaviour, which neither of us can predict with certainty.

Since this RFC would be a hardfork and quite a hefty change of the peercoin protocol, I would like to see a broader discussion with more participants. The initial comments from November do read a bit meager.

So far, the Peercoin blockchain has never been been attacked by malicious minters, despite the low participation.

Ötzi

@Nagalim any recent thoughts and updates on this idea?

The basic question here is whether we want to increase mint rate in the short term. This proposal cannot possibly result in fewer minters than the current state because it will bottom out to a 5x increase for a long time. The entire purpose of the proposal is to make such an increase in mint rate more palatable by providing a feedback mechanism to lower inflation above a minimum participation rate (20%). We are currently nowhere near this number. I think a more drastic base increase like 10% reward to 10% participation would cause exacerbation of attacks, but 5% should be fine.

The only changes I made were several months ago, where I added a recursive calculation to punish super big one-block minters, and acknowledged the seasonal timing attack. Ultimately, no timing attack will be functional at all below 20% participation and are essentially second-order behavioral changes compared to the baseline modification of striving to increase participation above 20% through increased mint inflation.

10.63% participation

2713253 PPC in Proof of Stake position across 190 wallets **, which is 10.63% of participation.

I believe 5% maximum reward cap should be increased slightly to 8-10% (1% global cap) should be a good incentive for larger and active wallets and idle balances on exchanges to enter minting, this way participation will quickly enter 30-50% as Minters are competing for rewards, while they need to seperate wallet into seperate nodes to unlock the maximum rate.

Effective Circulation

21.5M Circulation (- 3 year dormant *), 12.5% Participation rate
16.94M Circulation (- 1 year dormant *) 16% participation rate

Scenarios:

Hypothetically, following PPC balances enters minting at max return rate, here is the effective return rate to share 1% of global inflation (255000 PPC)

  • When 4.41M PPC locked for minting process, (let’s say No 1 wallet with 1.7M PPC resumes mining), max return 5.78%
  • When 7M PPC locked for minting process, max return 3.642%
  • When 10M PPC locked for minting process, max return 2.55%

As you can see, we can go slightly higher than 5% cap (5X) to 8% (8X) to improve the perception of Peercoin value proposition, but in practice incentives for participants will quickly balance out the network inflation.

References

Javascript

  var stakes = $('#pools-share tr[title]').map(function() { return this.getAttribute('title') })
  var ppcs = stakes.map(function(_, s) { var match = s.match(/(\d+(?:,\d+)) PPC/) ; return match && parseInt(match[1].replace(",","")) })
  Array.from(ppcs).reduce((a,b) => a+b, 0)

There are 2 reasons for the 5% cap as opposed to 10%.

  1. A 5% cap incentivizes a blanket increase in value until participation increases to 20%. As this is higher than the network has ever had it is essentially impossible for this proposal to result in a decrease in participation unless the change drove away interest for social reasons. If the rate reduced above 10% participation (as it would for a maxinflation of 10%), we would reach the situation that otzi is afraid of nearly instantaneously. While I don’t believe it would result in decreased participation, you could hypothetically imagine that a maxinflation of 10% could result in decreased participation compared to the present state. This causes it to be more contentious than it would otherwise need to be.

  2. A 10% block reward greatly exacerbates mint-domination attacks such as stake grind and N@S. A 5x reward rate is already pushing it in my opinion, and a 10x reward rate could easily result in there being a large reward for decreasing the participation of other minters. This is because the compounding interest becomes very important. In the current 1% paradigm, compounding interest is essentially negligible, so it is no big deal if your blocks are rejected, you can always just mint in the next window. With a 10% reward, the compounding interest becomes very powerful, and a n orphaned block becomes a real loss. I would suggest we keep the rate closer to the standard inflation rates of stable fiat currencies, which tends to be 1-5%.

As a thought experiment, you should consider a situation where there is no maxinflation, only the 1% global. In this experiment, rather than thinking about economic means, think only about the security implications of near-infinite instantaneous inflation for a solo-minter. Someone who forks and forges their own chain is rewarded greatly compared to those who participate with others. This is N@S, but there are other attacks this exacerbates as well. The 5% maxinflation was designed with these attacks in mind.

Also, I just want to clarify that at a >20% participation rate, the protocol would behave exactly the same whether maxinflation was 5%, 8%, or 10%. The only difference would be between 10-20% participation rate. However, some (most not, but some) of the attack vectors with a higher maxinflation would still be worse even at 30-50% participation, though there would be no benefit to the minters at that participation level because they would be operating below the maximum.

2 Likes

Price and market activity has increased since a month ago.

  • 3.298892/25.5 participation is almost at 13% now
  • 3.298892/21.5 (1y dormant) 15%
  • 3.298892/16.94 (3y dormant) 19.4%

Perhaps we can adjust the incentive slightly without run into the risks that you have covered.

Given the recent increase, I can see participation can easily reach 20% with a further increase in market activity.

Perhaps 20% participation rate can be increased slightly to allow the range to be 10-30%, whereas individual cap increase slightly to 8X instead of 5X?

Here is a list of POS/staked tokens with metrics on their participation rate, reward rate, etc.
https://staked.us/yields/

My intention is to understand if there are any economic changes that will increase the participation rate, the attractiveness of Peercoin from market participant’s perspective (crypto trader, investor or enthusiast )

The minimum participation and the cap are the same variable, if we fix the 1%/year inflation, as 20% times 5 equals 100%. Are you making the argument for more than 1%/year PoS inflation? I agree there are arguments to be made on that front, but it may be difficult to reach consensus on that point when also making a heavy protocol change like this.

Also, I’m not sure where you’re getting your participation numbers from, other than market participation. I tend to use the PoS difficulty as a benchmark for minting participation, which has been consistently around 7 for a while now. This proposal is more about minting participation than market participation, though of course I agree that increased yield will likely result in increased market participation.

I want to be clear here, I don’t mean to be a naysayer. I appreciate the desire to harness the active market participants as potential minters, and how that might benefit the ecosystem of the coin on many levels. My perspective comes from a confluence of forces pulling in several directions, and ultimately an attempt to achieve consensus by adhering to a rule of ‘if none of our core principals change, then no one will reject the proposal’. There are two core principals at play here. The first is the 1% total mint inflation, which the community has philosophically incorporated. It would feel amiss to alter this number because of a slippery-slope type feeling akin to the aversion toward changing Bitcoin’s ultimate supply cap. If we change it once, who’s to say we won’t change it again? The second principal is one of hard coded limits, which you can see in places such as the maximum rate of change of the mint difficulty. This point was pretty much the only comment I have attained from Sunny King when asked about this proposal, he mentioned that having a maxinflationadjustment was a beneficial concept as a protection for the protocol. Now, I will agree that the 5% personal inflation (which is consequentially accompanied by 20% minimum mint participation before the feedback loop is invoked, assuming the first principal of 1% total inflation) is somewhat arbitrary. However, I believe 5% interest is a critically incentivizing number, being larger than nearly all ‘guarenteed’ fiat interest rates, such as bond yields. The 20% mint participation number is coincidentally also a very important threshold historically, as it is larger than Peercoin has ever experienced for a prolonged period of time. Ultimately, I believe that the numbers in the proposal as they stand are very poignant as compromises for blanket consensus while still maintaining the drive towards a higher level of motivation for market participants to take that extra step to become minters. I truly do not mean to dismiss your analysis in any way, and am happy to continue discussion about the numbers chosen for this proposal.

2 Likes

Great to hear that you are open for discussion.

  1. The first is the 1% total mint inflation, I agree global inflation should not be changed.
  2. Participation POW mining still contribute to the majority of issuance. Let’s say at 5% participation then personal reward rate might be capped at 8% (8X), then it quickly scales down as more participant joins POS minting to balance out rate of reward and participation.

I am trying to propose is to allow these levers to be scalable between different factors like

A) coin price and
B) demand/supply market dynamic will play a role to incentivize
C) higher stake rate.

Given RFC11 is the first significant economic update to the network, there have been a lot more Proof-of-stake blockchains since 2012. This needs wider community feedback to consider.

  • Would higher POS participation make the network more secure and attractive without introducing significant risks?

  • How do we scale this reward rate while respecting 1% global inflation to improve Peercoin adoption?

The scenario: If personal reward cap scale down from 8 to 1%, as participation increase from 10-30%

I have a example of linear scaling between personal reward rate and participation.

This has been increasing in the last 3 month, at the current participation rate of 13.6% (3.479M/25.5M) the reward rate would be around 6.6%.

Feedbacks are welcome on what these levers should be, and how it should scale.

Generally higher PoS participation is correlated with higher security. More specifically, a higher mint difficulty corresponds to a higher security.

I would avoid coupling any protocol rules to the active (or perceived) market, that would be a recipe for failure.

This is possible and feasible to implement, but I would like to address the individual differences between your proposal and the proposal how it stands.

  1. You decrease the beginning threshold to 10%, allowing for a higher adjustmentmaximum, which you restrict at 8.

  2. You steepen the curve such that it reaches an adjustment of 1 at 30% participation instead of 100%.

I will respond to these changes with two points, which roughly correlate to the two changes.

  1. Lowering the beginning threshold and steepening the curve exacerbates the concerns Otzi had. This could result in some form of self-limiting or cyclical behavior wherein people attempt to game the system. It is important to note that there is a subtle but important difference between people claiming their inflation and people actively minting to increase the mint difficulty. One creates chain security while the other barely helps at all. We do not want to turn continuous minters into seasonal minters. I don’t believe the original proposal would to that, but an 8x slope over a 20% change in participation may cause that.

  2. The second point boils down to why leave money on the table? If you are bottoming at 30% participation, you are only awarding 0.3% total PoS inflation at that level. In fact, if we look at total amount of coins given out per year, you can see that it maxes out around 17.5% participation with ~0.94% inflation (aside from the full 1% given at 100% inflation). Game theory would state that the system will equilibrate around that point. Such an eventuality would clearly be less than ideal.

1 Like

Increased Load on Nodes

Any measurements on a stakebox for how much of an increased load this proposal would have? There’s something to be said for an energy efficient and sustainable blockchain .

When speaking with the devs, it was made clear that by simply indexing just one more variable during the initial sync, there will be only a trivial additional load. Thanks for bringing this to my attention, I should probably edit the proposal to reflect that.

1 Like

I have added the bit about trivial load, as well as generalized functional form and something about ‘futuristic’ minters. The futuristic minters section is speaking to the recent understandings of rfc2-4 about multisig minting and even mint pools. Rfc12 would be independent of this point.

It has come to my understanding that with the recalc to 1% inflation, peercoin will not become deflationary when lots of fees are burned. Just a important detail to consider…

We don’t want to become deflationary, it’s important to have some inflation in order to keep the economic system healthy.

While peerchemist is correct that the aim is not to become deflationary, the txn fees are not included directly in the calculation of adjustment (just like pow rewards are not directly included), so it is still entirely possible that some time in the future the txn fees overtakes the pos rewards.

This proposal was discussed recently in a team meeting. We will attempt to simplify implementation and will continue to discuss it.